Gene Expression Patterns Associated with Survival in Glioblastoma

Abstract

The aim of this study was to investigate gene expression alterations associated with overall survival (OS) in glioblastoma (GBM). Using the Nanostring nCounter platform, we identified four genes (COL1A2, IGFBP3, NGFR, and WIF1) that achieved statistical significance when comparing GBM with non-neoplastic brain tissue. The four genes were included in a multivariate Cox Proportional Hazard model, along with age, extent of resection, and O6-methylguanine-DNA methyltransferase (MGMT) promotor methylation, to create a unique glioblastoma prognostic index (GPI). The GPI score inversely correlated with survival: patient with a high GPI had a median OS of 7.5 months (18-month OS = 9.7%) whereas patients with a low GPI had a median OS of 20.1 months (18-month OS = 54.5%; log rank p-value = 0.004). The GPI score was then validated in 188 GBM patients from The Cancer Genome Atlas (TCGA) from a national data base; similarly, patients with a high GPI had a median OS of 10.5 months (18-month OS = 12.4%) versus 16.9 months (18-month OS = 41.5%) for low GPI (log rank p-value = 0.0003). We conclude that this novel mRNA-based prognostic index could be useful in classifying GBM patients into risk groups and refine prognosis estimates to better inform treatment decisions or stratification into clinical trials.